TL;DR
This paper investigates how approximations in modeling a patient's skull shape affect dose calculations for Gamma Knife radiosurgery. They found that while the standard 24-point skull model is generally accurate, significant discrepancies can arise in extreme cases, potentially leading to treatment time errors of up to ~5%. This highlights the trade-off between speed and precision in radiosurgery planning.
What Problem Does This Paper Address?
Gamma Knife radiosurgery uses a sophisticated system to deliver highly focused radiation to treat brain lesions. A crucial step in planning these treatments is accurately calculating the radiation dose delivered to the target and surrounding tissues. This calculation relies on a model of the patient's head, specifically the skull's shape, to determine how radiation beams are attenuated as they pass through. The Leksell GammaPlan (LGP) software, commonly used for this purpose, models the skull using only 24 discrete measurement points. The problem is that this simplified model might not perfectly capture the actual skull's complex geometry, especially in cases with unusual head shapes or after surgical interventions like craniotomies. These discrepancies between the modeled and actual skull could lead to inaccuracies in dose calculations, potentially impacting treatment effectiveness and safety. The authors aim to quantify the extent of these inaccuracies and their impact on treatment planning.
Methods
The researchers employed a multi-pronged approach to assess the impact of skull shape approximations:
- Discrepancy Simulations: They first used LGP to simulate treatment scenarios with intentionally introduced uniform and localized errors in the skull model. This involved adjusting the skull radius by 1 mm or 5 mm and simulating localized errors at various shot positions within the simulated head.
- Manual Skull Measurement Adjustment: In a clinical setting, they analyzed 24 patient cases where the standard 24-point skull model measurements were adjusted to better match the patient's actual skull contour as seen in CT or MR images. They then compared the treatment times calculated with the original and adjusted skull models.
- Image-Based Skull Contours: For a subset of cases, they went a step further by extracting the actual patient skull contour directly from CT or MR images. This image-based contour was then used in a custom-developed dose verification software to calculate treatment times, which were compared to those calculated using the adjusted LGP skull model.
Datasets
The study utilized data from two sources:
- Simulated Data: For discrepancy simulations, the researchers generated theoretical scenarios within the Leksell GammaPlan software.
- Patient Data: The core of the study involved analyzing 24 patient cases treated at the Winnipeg Centre for Gamma Knife Surgery. These cases represented a variety of treatment sites, including trigeminal neuralgia, acoustic neuroma, peripheral lesions, inferior lesions, multiple metastases, and a rare squamous cell scalp lesion. For these patients, both MR and CT scans were available.
Key Insights and Ideas
- Centrally Located Shots are Robust: The study revealed that shots located near the center of the head are relatively insensitive to skull shape approximations. The effects of localized over- or underestimations in the skull model tend to cancel each other out in these central regions.
- Peripheral Shots are More Sensitive: As shots move towards the periphery of the head, the impact of skull model discrepancies becomes more pronounced. Beams passing through inaccurately modeled regions of the skull can lead to larger errors in dose calculation.
- The 24-Point Model is Generally Sufficient: For most clinical scenarios, the standard 24-point skull model in LGP provides a quick and convenient way to approximate skull shape without introducing significant errors in treatment time.
- Extreme Cases Matter: The research highlights that in certain "extreme cases," such as those involving peripheral shots or highly irregular skull shapes (like after a craniotomy or in the case of the squamous cell scalp lesion), the 24-point model can lead to substantial discrepancies, potentially resulting in treatment time errors approaching 5%.
- Image-Based Contours Offer Higher Accuracy: Using actual patient contours extracted from CT or MR images, as opposed to the LGP's approximated model, can lead to more accurate dose calculations, especially in complex cases.
Key Results
- Uniform Errors: A uniform 1 mm or 5 mm error in all skull measurements affected treatment time by 1.5% and 3.4%, respectively, for centrally located shots.
- Localized Errors: The error in treatment time due to localized skull discrepancies ranged from less than 0.5% to about 3.0%, depending on the shot location.
- Manual Adjustment Impact: Adjusting the 24-point skull measurements to better match imaging data resulted in only small changes in treatment times, with average differences generally below 0.5% for most treatment sites.
- Image-Based Contour Impact: Comparing treatment times between the adjusted LGP model and image-based contours revealed larger discrepancies. For the squamous cell scalp lesion, an error of 4.1% was observed, demonstrating the limitations of the 24-point model in extreme situations.
Limitations and Future Work
- Limitations of the 24-Point Model: The authors acknowledge that the 24-point model is inherently limited in its ability to capture complex skull shapes, particularly in cases with significant geometric variations or after surgery.
- Future Software Enhancements: The paper suggests that future versions of LGP could benefit from an option to directly use actual patient contours from CT or MR data. Alternatively, incorporating a larger number of model points could improve accuracy.
- Patient Inhomogeneities: The study also points out that LGP assumes uniform tissue density and does not account for attenuation differences due to bone, air cavities, or interfaces. Addressing these factors would require more sophisticated dose calculation algorithms and potentially the routine collection of CT imaging data.
- Further Investigation: The authors recommend further investigation into the impact of patient inhomogeneities and the development of more advanced dose calculation methods to improve accuracy in complex radiosurgery cases.
Summary updated